Pair Trading Strategies Using Machine Learning: A Case of PSX Firms
Pakistan Business Review · 2021
DOI: 10.22555/pbr.v22i3.179 ↗Kurz gefasst
Fachartikel in der Zeitschrift Pakistan Business Review, 2021. Verfasst von Muhammad Khalid Sohail. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Pair trading is generally known as a profitable strategy of an investment. Here, in this study DBSCAN clustering algorithm (machine learning) in addition to the traditional pair trading technique. is used. By using this algorithm 3 cluster are identified from EPS, Market-Cap, sector classification and BVS (fundamental variables) with other factors formed from PCA on the returns of daily data of two years of the sample firms. Pairs are also formed based on traditional distance approach of Gatev et al. (2006). Sample consists of 80 stocks from five different sectors; banking, chemicals, cement, textile and food & care products from year 2011 to 2019.Under the machine learning a remarkable 1.16% average excess monthly return with Sharpe ratio of 2.48 is to be observed. For risk adjusted returns, Jensen’s alpha under CAPM is also to be observed positive and significant. The results also authenticate mean revision and market neutrality at PSX. Investors can get positive returns through pair trading at PSX. Portfolio and fund managers can form the pair trading strategy to reap the profitability for their clients and specially they can get higher returns by using machine learning approach.
Zitieren
APA
Sohail, M. K. (2021). Pair Trading Strategies Using Machine Learning: A Case of PSX Firms. Pakistan Business Review. https://doi.org/10.22555/pbr.v22i3.179
Vancouver
Sohail MK. Pair Trading Strategies Using Machine Learning: A Case of PSX Firms. Pakistan Business Review. 2021. doi:10.22555/pbr.v22i3.179
BibTeX
@article{Sohail2021,
author = {Sohail, Muhammad Khalid},
title = {Pair Trading Strategies Using Machine Learning: A Case of PSX Firms},
journal = {Pakistan Business Review},
year = {2021},
doi = {10.22555/pbr.v22i3.179}
}
EUVIMED